Personal Identification and Description 191 



sets agreed in all their three couplets of fingers; four sets in two, and 

 five sets in one of their couplets. There are instances of partial agreement 

 in five others, and only complete disagreement in one. Of the second series 

 of 17 twins Galton contents himself by saying that two sets agreed in two of 

 their couplets and five agreed in one, without giving details. He concludes that : 



" there cannot be the slightest doubt as to the strong tendency to resemblance in the finger 

 patterns of twins." (p. 186.) 



Unfortunately Galton gives no measure of the probability of the random 

 occurrence of similar resemblances, and we are unable to compare what is 

 the relative degree of resemblance of twins and ordinary siblings. 



Perhaps the best appreciation the reader can rapidly form of the degree 

 of resemblance in the finger-prints of like twins can be obtained by carefully 

 examining our Plate XVIII which gives the finger-prints of a pair of like 

 twins from the Galtoniana. 



The last problem Galton touches on is that of parental heredity. Here 

 he has only 27 pairs of parents, whom he chooses because on one of the three 

 fingers, fore, middle, or ring, they have the same pattern. He has 4 cases of 

 the forefinger, 14 of the middle finger and 9 of the ring finger. These 27 

 pairs of parents have 44 sons and 65 daughters; 22 out of the 44 sons, 37 

 out of the 65 daughters have the same pattern on the same finger as their 

 parents. In 19 cases out of the 27 both parents had loops of type No 42, and 

 in 48 cases out of their 75 children there was also a loop on the same finger; 

 that is to say, in about 64 °/„ of cases, while the normal percentage is about 

 33 °/ - Thus, according to Galton's method, the resemblance is about 48°. 

 This seems to show a much greater value for filial resemblance in looping 

 than had been found for fraternal resemblance. Yet in analysing these 

 parental sets, Galton is rather apt to desert the method he adopted 

 for fraternal resemblances, namely, of terming two points like or unlike 

 according as they are of the same or not the same pattern in his C-set of 

 53 patterns. Thus he has 3 parental sets with No. 14 tendrilled loops; they 

 have 17 children of whom only 3 have No. 14 pattern; he says, however, that 

 No. 14 counts as a whorl, and that the 17 have 11 whorls and only 6 loops. 

 Few, however, of the remaining 8 whorls bear close resemblance to No. 14. 

 Galton gives no general measurement of parental heredity. 



This raises, indeed, the broad question whether it is really the pattern 

 which is inherited, or merely a tendency to arch, to loop, or to whorl with- 

 out regard to the individual character of the pattern. Galton remarks 

 (p. 187) that the finger-prints of twins while tending to be of the same 

 pattern, cannot be mistaken one for the other; in other words, the number of 

 ridges and the minutiae differ*. Thence he leads us to a very fertile suggestion, 

 which neither he nor anyone else later, so far as I know, has ever worked out : 



" It may be mentioned that I have an inquiry in view, which has not yet been fairly 

 begun, owing to the want of sufficient data, namely to determine the minutest biological unit 

 that may be hereditarily transmissible. The minutiae in the finger-prints of twins seem 

 suitable objects for the purpose." (p. 187.) 



* Our Plate XVIII suggests that Galton in this statement has somewhat over-emphasised 

 the divergence between the finger-prints of twins. 



